Two-Handed Input Using a PDA and a Mouse

Two-Handed Input Using a PDA And a Mouse

Brad A. Myers, Kin Pou (“Leo”) Lie , and Bo-Chieh (“Jerry”) Yang Human Computer Interaction Institute School of Computer Science Carnegie Mellon University Pittsburgh, PA 15213 [email protected] http://www.cs.cmu.edu/~pebbles

ABSTRACT around that could serve as an extra input device for the computer: their Personal Digital Assistant (PDA). PDAs, We performed several experiments using a Personal Digi- such as 3Com’s Palm Pilots and Microsoft’s Windows CE tal Assistant (PDA) as an input device in the non- devices, are designed to be easily connected to PCs and dominant hand along with a mouse in the dominant hand. have a touch-sensitive screen which can be used for input A PDA is a small hand-held palm-size computer like a and output. Furthermore, newer PDAs, such as the Palm V 3Com Palm Pilot or a Windows CE device. These are be- and the HP Jornada 420, are rechargeable, so they are sup- coming widely available and are easily connected to a PC. posed to be put in their cradles next to a PC when the user Results of our experiments indicate that people can accu- is in the office. Therefore, if using a PDA in the non- rately and quickly select among a small numbers of dominant hand proves useful and effective, it should be buttons on the PDA using the left hand without looking, increasingly easy and sensible to deploy and configure us- and that, as predicted, performance does decrease as the ing hardware devices that users already have. number of buttons increases. Homing times to move both hands between the keyboard and devices are only about Another advantage of PDAs over the input devices studied 10% to 15% slower than times to move a single hand to in previous experiments is that they are much more flexi- the mouse, suggesting that acquiring two devices does not ble. PDAs have a display on which virtual buttons, knobs cause a large penalty. In an application task, we found that and sliders can be displayed, and they can be programmed scrolling web pages using buttons or a scroller on the PDA to respond to a wide variety of behaviors that can be well- matched the speed of using a mouse with a conventional matched to particular tasks. However, a disadvantage is scroll bar, and beat the best two-handed times reported in that the controls on the PDA screen are virtual, so users an earlier experiment. These results will help make two- cannot find them by feel. Research is therefore needed to handed interactions with computers more widely available assess how well the PDA screen can work as a replacement and more effective. for other input devices that have been studied for the left hand. Keywords: Personal Digital Assistants (PDAs), Hand-held computers, Palm Pilot, Windows CE, Two-Handed Input, This paper reports on several experiments that measure Smart Environments, Ubiquitous Computing. various aspects of using a PDA as an input device in the non-dominant hand. Two experiments are new and are de- INTRODUCTION signed to measure the parameters of using a PDA. One experiment repeats an earlier study but uses a PDA in the Many studies of two-handed input for computers have of- non-dominant hand. Since the actual pragmatics of input ten shown advantages for various tasks [1, 3, 7, 9, 15]. devices can have a large impact on their effectiveness [2, However, people rarely have the option of using more than 8], we wanted to determine whether the results seen in just a mouse and keyboard because other input devices are prior experiments would also apply to using PDAs. relatively expensive, awkward to set up, and few applications can take advantage of them. However, increasing In summary, the results are: numbers of people now do have a device that they carry x People can quickly and reliably hit large buttons drawn on the PDA with their left hands without looking. 99% of the button taps were correct on buttons that are 1-inch Submitted for Publication square in a 2x2 arrangement. With a larger number of smaller buttons, the accuracy significantly decreases: 95% were correct for 16 buttons that are ½ inch on a side arranged 4x4. The time from stimulus to button tap Two-Handed Input Using a PDA And a Mouse - 2 - ** SUBMITTED FOR PUBLICATION** was about 0.7 sec for the large buttons and 0.9 seconds scrolling task using a variety of devices, which is a repeat for the smaller buttons. of an earlier experiment [15]. The subjects reported a x In a task where the subjects had to move both hands number of problems with the scrolling devices on the PDA from the keyboard to the PDA and the mouse and then in the last task, so we redesigned the scrolling devices, and back, we found that it took an average of 0.791 seconds in a second study with new subjects, we evaluated the per- to move both hands from the devices to the keyboard. formance of the new scrollers on the same task. Each of This was about 13% longer than moving one hand from these experiments is described below. the mouse to the keyboard (which took 0.701 sec). Apparatus Moving to a PDA and mouse from the keyboard took an average of 0.838 seconds, which is about 15% longer Subjects sat at a normal PC desktop computer that was than moving one hand to the mouse (0.728 seconds). running Windows NT. On the right of the keyboard was a x In a repeat of the experiment reported in [15], subjects mouse on a mouse pad. On the left was an IBM WorkPad were able to scroll large web pages and then select a link 8602-30X PDA (which is the same as a Palm IIIx). In the on the page at about the same speed using buttons or a first study, we put the WorkPad in its cradle. Subjects scroller on a PDA compared to using the mouse with a complained that the WorkPad was wobbly it its cradle, so conventional scroll bar. The times we found for scrolling for the second study, the new subjects used a WorkPad with buttons on the PDA were faster than any of the resting on a book and connected by a serial cable to the times in the earlier study, including the 2-handed ones. PC. There were no further comments about the positioning. RELATED WORK The WorkPad has a 3¼-inch diagonal LCD display screen There have been many studies of using input devices for (about 2 ¼ inches on a side) which is touch sensitive. It is computers in both hands, but none have tested PDAs in the 160x160 pixels. Figure 1 shows a picture of the full Work- left hand, and we were unable to find measurements of Pad. homing times from the keyboard to devices for two-handed The software running on the WorkPad was the Shortcutter use. program [10] that allows panels of controls to be created so One of the earliest experiments measured the use of two that each control sends specified events to the PC. The hands in a positioning and scaling task and for scrolling to software on the PC consisted of various applications spe- known parts of a document [3]. This study found that peo- cifically created for this experiment (except in the scrolling ple naturally adopted parallel use of both hands and could task, which used the Netscape browser running a custom scroll faster with the left hand. Theoretical studies show JavaScript program to collect the data). that people naturally assign different tasks to each hand, and that the non-dominant hand can support the task of Typing Test the dominant hand [6]. This has motivated two-handed We used a computerized typing test called “Speed Typing interfaces where the non-dominant hand plays a support- Test 1.0” [14]. The subjects were asked to type a para- ing role, such as controlling other drawing tools [9] and graph displayed on the screen as quickly as possible. adjusting translation and scaling [3, 15]. Other studies have tested two-handed use for 3D interaction [1, 7] and Button Size Task found the extra input to be useful. In this task, the PDA displayed between 4 and 16 buttons There has been prior work on using PDAs at the same time in eight different arrangements: 2 rows by 2 columns, 2x3, as regular computers for various tasks including meeting 3x2, 2x4, 4x2, 3x4, 4x3, and 4x4 (see Figure 1). To con- support [11], sharing information [12], and to help indi- trol for ordering effects, half of the subjects used the order viduals at their desks [10], but we found no prior work on shown above, and the other half used the reverse order measuring performance of non-dominant hand use of (4x4 first down to 2x2 last). In the 2x2 condition, the but- PDAs. tons were about one inch square, and in the 4x4, they were about ½ inch square. EXPERIMENTAL DESIGN At the beginning of each condition, a picture was dis- Two new studies were performed. In the first, the subjects played on the PC screen showing the corresponding layout did five tasks in a row. The first task was a typing test to of the buttons (with the same size as the PDA). Then one see how fast the subjects could type. Next, they performed of the buttons was shaded black (see Figure 2). The sub- a button size task to measure the error rates and speeds jects were asked to tap on the corresponding button on the when tapping on different size buttons on the PDA. Next, WorkPad as quickly and accurately as possible with a fin- the subjects performed a homing speed task where we ger on their left hand. The stimulus button was then measured the how quickly the subjects moved among the cleared on the PC and the next stimulus button was shaded keyboard and the devices. Finally, they performed a 500 milliseconds later. The stimuli appeared in random Two-Handed Input Using a PDA And a Mouse - 3 - ** SUBMITTED FOR PUBLICATION** order. A total of 48 stimuli were used in each condition. and they start and end with keys that are under the oppo- Every button appeared the same number of times. For ex- site hands. The user could not exit the field until the word ample, for the layout of 2 rows by 2 columns, each button was typed correctly. After typing the word correctly into appeared 12 times, while for the layout of 3 rows by 4 col- the textbox, the subject then continued to perform the same umns, each button appeared 4 times. There was a break selection-typing operation in the next textbox. The trial after each condition. Our hypotheses were that people ended when all 14 textboxes on the screen were filled in. could more accurately select among fewer, larger buttons, There was a break after each trial. We measured the time and that people could make selections without looking at from the mouse and WorkPad click to the first character the WorkPad. typed, and from the last character typed to the first movement of the mouse or tap on the WorkPad. We did not count the time spent actually typing, and we eliminated the times for the first and last words, because they were biased by start-up and transients. We hypothesized that moving to the WorkPad and the mouse would not take much longer than moving one hand since people would move both hands at the same time. We were also interested in the actual numbers for the time measurements. These might be used with future models of human performance for two-handed homing tasks.

Figure 1. Left: a picture of a Palm Pilot (the WorkPad is similar) showing the 2x2 layout of buttons. Right: the screens for 3x2, 2x3, 4x3 and 4x4. The other layouts are similar.

(a) (b) (c) Figure 3. (a) The button scroller on the WorkPad used in the first experiment. (b) The Slide Scroller and (c) Rate scroller used in both experiments. Figure 2. Part of the PC screen showing the stimulus during the 4x3 condition of the button task. Scrolling Task For this task, we were able to replicate the conditions of a Homing Speed Task previous experiment [15] exactly.1 The purpose of this task The purpose of this task was to measure the times to move was to evaluate and compare subjects’ performance in the hands back and forth from the keyboard to the mouse scrolling web pages in a standard browser using different and WorkPad as the subjects switch between two-handed scrolling methods. The web pages contain text from an selection operation and two-handed typing. We compared IBM computing terminology dictionary, and each page is moving a single hand to and from the keyboard to moving about 12 screen-fulls of text. In each web page a hyperlink both hands. with the word “Next” is embedded at an unpredictable lo- cation. The subjects were asked to find the target hyperlink There were three conditions with three trials in each. In by scrolling the web page using the different scrolling each trial, 14 textboxes were shown on the screen with a mechanisms. Once the link was visible, they used the label in front of each. The conditions were that the subjects mouse in the conventional way to click on it. Clicking on had to first select a text box by either clicking in the field the hyperlink brought the subject to the next web page. For with the mouse in the usual way, tapping on a full-screen each condition, the subjects first performed a practice run button on the WorkPad (which therefore worked like a of 10 pages, during which they were asked to try out the “TAB” key and caused the cursor to jump to the next scrolling method without being timed. Then, the subjects field), or tap on the WorkPad and click the mouse at the did two consecutive trials of 10 pages each as fast as they same time. In other words, the selection operation in this could. last condition was like a “Shift-Click” operation in which the button on the WorkPad was treated as a Shift key. Af- ter the textbox was selected, the subjects typed the word indicated on the left of the textbox. The word was either “apple” or “peach” (in alternating order). These words 1 Thanks very much to Shumin Zhai of IBM for supplying the experimental were chosen because they are easy to type and remember, material from the earlier study. Two-Handed Input Using a PDA And a Mouse - 4 - ** SUBMITTED FOR PUBLICATION** buttons were not needed—see Figure 4a). We also tried to improve the rate scroller, by adjusting the scroll speeds and the areas where each speed was in affect. Finally, we added a new (sixth) condition: x Scrolling using an “absolute scroller,” where the length of the scroller represented the entire document, so put- (a) (b) ting a finger at the top jumped to the top of the Figure 4. (a) The revised button scroller on the WorkPad used document, and the bottom represented the bottom (see in the second experiment. (b) The “absolute scroller”. Figure 4b). The user could also drag up and down to scroll continuously. Therefore, it was as if the scroll The condition with the fastest time in the previous experi- bar’s indicator was attached to the finger. The motiva- ment used a “pointing stick” joystick to scroll, but we were tion for this scroller was that we noticed that most not able to reproduce this condition.2 The conditions we people in the mouse condition of the first session used in our first experiment were: dragged the indicator of the scroll bar up and down, and we wanted to provide an equivalent WorkPad scroller. x Scrolling using the mouse and the regular scroll bar. x Scrolling using a “scroll wheel” mounted in the center Subjects of the mouse (a Microsoft “IntelliMouse”). We were There were 12 subjects in the first study, which took about careful to explain to the subjects the three different ways an hour and they were paid $15 for participating. 12 dif- the wheel can be used, including rolling the wheel, or ferent subjects did the second study, which took about ½ tapping or holding the wheel down to go into “scroll hour and they were paid $10. All subjects were Carnegie mode” where the further you move the mouse from the Mellon University students, faculty, or staff. 25% (6 out of tap point, the faster the text scrolls. The subjects could 24) were female, and the age ranged from 19 to 46 with a choose which methods to use. median of 26. All were moderately to highly proficient x Scrolling using buttons on the WorkPad (see Figure 3a). with computers, and half had used PDAs. The data from There were 6 buttons that scrolled up and down a line, some extra subjects were eliminated due to technical diffi- up and down a page, and left and right (which were not culties. The measures from two subjects who were left- needed for this experiment). The buttons auto-repeated handed are not included in the data, but informally, their if held down. numbers did not look different. x Scrolling using a “slider” on the WorkPad (see Figure RESULTS 3b). Putting a finger on the slider and moving up or down moved the text the corresponding amount. If you General reach the edge of the slider, then you need to lift your Pearson product-moment correlation coefficient between finger and re-stroke. Tapping on the slider has no effect typing speed and tap speed in the button size task (namely since only relative movements are used. the mean tap speed across all 8 layouts) was .60, which x Scrolling using a “rate scroller,” which acted like a rate- means the faster typists were somewhat faster at tapping. controlled joystick with three speeds (see Figure 3c). The correlation coefficient between typing speed and the Putting a finger on the WorkPad and moving up or speed for moving one hand to the keyboard in the homing down started the text moving in that direction, and task was .79 which means, as expected, subjects who were moving the finger further from the start point scrolled better typists could put their hands in the home position faster. more quickly. There was little correlation of typing speed to the other measures in the homing task. The correlation The order of the conditions was varied across subjects. coefficient between typing speed and scrolling speed (in the revised scrolling task) across all 6 conditions and both Revised Scrolling Task trials was 0.32, which means there was little correlation We received a number of complaints and suggestions about for the scrolling task. the scrollers on the WorkPad in the first session, so we redesigned some of them and repeated the scrolling task in a Age and gender did not affect the measures. second study with new subjects. In this study, we only used four buttons for the button scroller (since the left and right Button Size Task Figure 5 shows the times to tap on the button measured from the time the stimulus appeared on the PC monitor. 2 We did not have a pointing stick to test, and anyway, it would have been These numbers only include correct taps. There were two difficult to connect one to the computers we had, which illustrates one of the orders for the trials, so each condition was seen by some claims of this paper—it can be difficult to connect multiple conventional input devices to today’s computers. Since the experimental set up was identi- subjects early in the experiment, and by other subjects cal to the original experiment [15], it should be valid to compare our times later. The chart presents the data for the early and late with the times reported there. cases along with the average of both. Two-Handed Input Using a PDA And a Mouse - 4 - ** SUBMITTED FOR PUBLICATION** ms ec Figure 7 shows the error rates for the various configura- 1000

900 867 790

775 tions, which varies from 1.04% to 4.17% for the subjects

800 751 712

700 664 616 600 593 who saw each condition later. The error rates do not differ 500 400 significantly among different layouts (F7,70=1.6, p=.14) nor 300 200 among different numbers of buttons (F4,40=2.1, p=.07). For 100 u 0 the 4 4 layout, 45% of the errors were in the wrong row, 2x2 2x3 3x2 2x4 4x2 3x4 4x3 4x4 48% were in the wrong column, and 7% were wrong in Ear ly Late Averag e both (on the diagonal from the correct button). There was no consistent pattern of where the problematic buttons Figure 5. Times to tap each button depend on the size. The were located (see Figure 8). times are shown for the subjects who saw each condition later. percent 2x2 (first) 2x2 (last) msec msec 100 90 2000 2000 80 70 1500 1500 60 50 1000 1000 40 30 500 500 20

10 4.17 4.17 2.08 2.43 2.08 2.43 1.04 0 0 0 1.04 0 8 16 24 32 40 4 0 8 16 24 32 40 48 2x2 2x3 3x2 2x4 4x2 3x4 4x3 4x4 Figure 6. Plot of all times for the 2x2 layout shows (on the left) Early Late Average learning happening for those subjects who saw this condition first, but not (on the right) for those who saw it last. Figure 7. Error rates for each condition of the button task. Numbers shown are for the subjects who saw each later. Figure 6 shows the times to tap on a button in the 2x2 trial for each of the buttons for each of the subjects. The left 18.5% Key: 11.11% graph is of those subjects who saw the 2x2 condition first, Error 8.33% and roughly matches the power law of practice. However, Rates: 5.56% for those subjects who did the 2x2 condition last, there was 2.78% no apparent learning during that trial, and the times are 0.0% flat. Therefore, we feel it is more valid to use the times Figure 8. Percent of the taps in each button that were in error in from only the subjects who saw the condition later. The the 4x4 layout. average time for just the second set is 593msec. Homing Speed Task As shown in Figure 5, and predicted by Fitts’s law [5, p. Figure 9 shows the times for moving each hand in the 55], the time to tap on a button is inversely proportional to various conditions of the homing speed task. When mov- the size of the button, ranging from 593 msec in the 2x2 ing only one hand at a time (top 4 rows), the subjects took condition to 867 msec in the 4x4 (for those the subjects 728 msec to move to the mouse and 701 msec to move who saw each condition later). back to the keyboard from the mouse. The times to move to The times to tap differ significantly among different num- the PDA were 744 msec to the PDA and 639 back. bers of buttons (F4,40=16.8, p<.001). There is significant When required to move both hands, the subjects took only interaction between button number and order of conditions slightly longer, requiring about 15% more time to acquire (2x2->4x4 or 4x4->2x2) (F4,40=6.0, p=.001), but the both the PDA and the mouse (838msec), and about 12% learning effect is most prominent among layouts with more time to acquire the keyboard (791 msec). small number of buttons. The Tukey Test at .05 significant level indicates that there is no significant difference be- 1H Keyboard->Mouse 728 tween the 4-button condition and the 6-button condition, 1H Keyboard->PDA 744 1H Mouse->Keyboard 701 between the 6-button and 8-button, or between the 8-button 1H PDA->Keyboard 639 and 12-button. However, the 12-button condition is faster Keyboard -> Mouse&PDA 838 15.1% than the 16-button condition by a statistically significant Mouse&PDA -> Keyboard 791 12.8% margin. Figure 9. Times in milliseconds to move hands. “1H” means when only one hand is moving. The third column shows the per- The times for different layouts of the same number of but- cent slowdown of moving both hands compared to the tons is not statistically significant, however: the Tukey corresponding one-handed mouse time. Test at .05 significant level indicates that times for the 2u3 are not statistically different from 3u2, 2u4 compares to Scrolling Task 4u2, and 3u4 compares to 4u3. As in the study we reproduced [15], the time for the first trial with each input device was for practice, so Figure 10 shows the times for the second and third trials. Two-Handed Input Using a PDA And a Mouse - 6 - ** SUBMITTED FOR PUBLICATION** sec difference between ratings of scroll wheel and button 100 89.4 scroller was not significant. The differences of ratings 90 75.9 78.7 80 70.0 among the three Pebbles scrollers were not significant. 62.965.0 70 60.7 59.761.3 60 53.2 50 Revised Scrolling Task 40 We were not happy with the performance of the scrollers 30 20 on the PDA, and the subjects provided useful feedback on 10 ways to improve them. Therefore, we performed iterative 0 design on the software, and tried the scrolling task again Trial 2 Trial 3 with 12 new subjects. Figure 12 shows that we were able to Mouse Scroll Wheel Button ScrollerS lide ScrollerRate Scrolle improve the performance of the new versions of the button scroller, but the rate scroller may be worse. The new ab- Figure 10. Times in seconds to scroll through 10 pages in trials solute scroller was quite fast. The ratings of the new 2 and 3 of the first version of the web page scrolling task using versions are shown in Figure 13 and parallel the perform- different input devices. ance.

great 3 sec 85.6 90 very good 2 1.7 78.7 1.3 80 69.4 70 61.6 good 1 57.8 0.4 55.6 54.2 0.0 60 50.3 49.9 51.6 46.8 44.9 OK 0 50 Mouse ScrollWheel Button Scroller Slide Scroller Rate Scroller -0.3 40 poor -1 30 20 very poor -2 10 0 terrible -3 Trial 2 Trial 3 Figure 11. Ratings of the various input methods by the subjects in the first version of the scrolling experiment. We used the same Mouse S cr oll Wheel B uttonS croller SlideScrollerRateScroller AbsS crolle scale as [15]. Figure 12. Times in seconds to scroll through 10 pages in trials 2 and 3 of the second version of the web page scrolling task. A repeated measure variance analysis shows that subjects’ great 3 completion time was significantly affected by input method (F = 13.3, p < .001). Trial 3 was significantly faster very good 2 1.6 4,44 1.1 0.8 than Trial 2 (F1,11=17.2, p<.001), showing a learning ef- good 1 0.5 fect. However, this improvement did not alter the relative OK 0 performance pattern of the input method (F4,44=.5, p=.73). M ous e Scr ol l Wheel B utton Sl ide Scr oll erRate ScrollerAbs-0.1 Scroller poor -1 Scr oller Taking the Mouse condition as the reference and averag- -1.1 ing over both trials, the scroll wheel, the Slide Scroller, very poor -2 and the Rate Scroller conditions were 28, 11, and 48 per- terrible -3 cent slower. The Tukey Test at .05 significant level indicates that the difference between mouse and scroll Figure 13. Ratings of the various input methods by the subjects wheel conditions, between the mouse and button scroller, in the second version of the scrolling experiment. and between the mouse and slide scroller conditions were A repeated measure variance analysis showed that subjects not significant, while the difference between mouse and completion time was significantly affected by input method rate scroller conditions was significant. (F5,55 = 29.3, p < .001). Taking the Mouse condition as the Figure 11 shows the subjects’ ratings of the various scrol- reference and averaging over both trials, the button scrollers using a rating scale from Zhai et. al. [15]. Contrary to ler was 8 percent faster but the Tukey Test at .05 the results of that previous study, the Tukey Test at .05 significant level indicates that such difference is not sig- significant level indicates that the difference between rat- nificant. The scroll wheel, the absolute scroller, the slide ings of mouse and scroll wheel was not significant. scroller, and the rate scroller conditions were 31, 7, 12, Subjects gave the mouse a significantly higher rating than and 64 percent slower than the standard mouse condition. the slide scroller, while the difference between ratings of The Tukey Test at .05 significant level indicates that the mouse and button scroller and the difference between rat- difference between mouse and scroll wheel conditions and ings of between mouse and rate scroller were not the difference between mouse and rate scroller conditions significant.. Subjects gave the scroll wheel a significantly were significant, while the difference between mouse and higher rating than slide scroller and rate scroller, while the absolute scroller conditions and between Mouese and slide scroller conditions were not significant. Two-Handed Input Using a PDA And a Mouse - 7 - ** SUBMITTED FOR PUBLICATION** DISCUSSION 0.3 seconds, so this might be subtracted from our measured time to get the predicted 0.4 seconds. Button Size The subjects were able to hit buttons quite accurately with An important observation is that, as predicted, subjects their left hand, especially for small numbers of buttons. moved both hands simultaneously, and this did not penal- The predicted decrease in performance with decreased ize the movement time much. The sum of the one-handed button size was observed. There seems to be a threshold of times to move from mouse and PDA to the keyboard is about 12 buttons before there is any affect due to the size. 1340msec (701+639). This is much larger than the time to move from both mouse and PDA to the keyboard in the We believe that we achieved expert performance (the two handed case which is 791 msec (1340msec is 69% learning curve flattened out) by the end of the experiment, larger). A similar relationship holds for the movement so we tried using the times in models of expert human from the keyboard to the PDA and mouse performance. One candidate is Fitts’s law, but we do not (728+7443=1473 > 838; 76% larger). know exactly where the subjects’ fingers were when they started to move to tap. Assuming a movement of about 2 Overall, it takes only about 15% longer to acquire both the inches and a target size of 1 inch (in the 2x2 case), Fitts’s mouse and the PDA than just to acquire the mouse, and it law as formulated in [5, p. 55] predicts a movement time takes only about 13% longer to get back to the keyboard of about 150msec, compared to our measurement of from both devices than from just the mouse. 593msec. In our task, however, there is also perception We were not able to find any prior studies of the time to and thinking time. For the smaller buttons (½ inch in the acquire two devices at the same time. Most studies of two- 16x16 case), Fitts’s law predicts an increase in time of handed use of input devices (including our button-size and about 100msec, but we saw an increase of about 275msec. scrolling tasks) allow the subjects to stay homed on the de- We observed that subjects looked back and forth from the vices. We found that moving both hands slowed down each monitor to the PDA, at an increasing rate depending on hand a little, but there was substantial parallel movement. the number of buttons to choose from. Therefore, we be- Realistic tasks are likely to include a mix of keyboard and lieve the performance cannot be modeled simply as a other input device use, so homing issues may be important. Fitts’s law task, but we were unable to find an appropriate alternative model. Scrolling Our results showing that users can tap up to 12 buttons ac- Our measured times for scrolling the web pages with the curately and quickly with the left hand is relevant since mouse (about 60 seconds) is a little faster than the time there are a number of applications where having several reported in [15], and in the revised web task, the time for buttons on the PDA would be useful. Examples include scrolling with the button scroller is 45.9 sec (average of scrolling with buttons (Figure 3a and Figure 4a), and pan- trial 2 and trial 3) which is faster than the time for scroll- els created with the Shortcutter tool for controlling a ing with the in-keyboard isometric joystick (around 50 compiler, playing music on the PC, reading mail, etc. [10]. sec). This shows that using the PDA can match or beat the speed of other non-dominant hand devices. Homing Times An interesting comparison is between their joystick, our Our one-handed homing time to move from the mouse to Rate Scroller (Figure 3c) and the scroll wheel used in its the keyboard (701 msec – see Figure 9) is longer than the most popular manner as a rate-controlled scroller. All pro- time to move from the PDA to the keyboard (639 msec). vide the same rate-controlled style of scrolling, but they This may be because the physical distance to the mouse have significantly different performances and ratings by from the home position on the keyboard is longer (14 users. Our attempt to improve the rate scroller obviously inches compared to 7 inches) due to the number-pad and did not help, showing that further work is needed to make arrow keys sections of the PC keyboard. In the other direc- this scrolling method effective. We observed that the fast tion, the increased time to acquire the PDA may be due to speed was much too fast, but the medium speed was too the unfamiliarity of homing to this kind of device. slow. The popularity of the scroll wheel and the success of In the classic study of text selection devices [5, p. 237], the the pointing stick give us reason to keep trying. Further- homing time to move from the space bar to the mouse was more, IBM did significant experimentation and measured as 0.36 seconds. This was measured from adjustments before the pointing stick had acceptable per- videotapes of subjects moving. An average homing time of formance [13]. Therefore, an important conclusion from 0.4 seconds was incorporated into the Keystroke Level the scrolling experiment is that the specific design and Model [4]. However, we measured one-handed homing pragmatics of the input methods has a very important in- times of around 0.7 seconds, which is substantially longer. fluence on the performance. Our time was measured from the time of the mouse click Another interesting result is that our subjects quite liked to the time that the first keystroke was recognized. Our the scroll wheel (average rating of 1.7 | very good), typing test shows that the average time per keystroke was whereas in the earlier study it was rated much worse (-1 | Two-Handed Input Using a PDA And a Mouse - 8 - ** SUBMITTED FOR PUBLICATION** poor) [15]. This may be due to the increased experience Citation of manufacturer’s or trade names does not constitute an official en- people have with a scroll wheel (many of our subjects have dorsement or approval of the use thereof. a scroll wheel on their own mouse), and because most of REFERENCES our subjects used it in its rate-controlled joystick mode, 1. Balakrishnan, R. and Kurtenbach, G. “Exploring Bimanual whereas most of the earlier study’s subjects used the roll- Camera Control and Object Manipulation in 3D Graphics Inter- ing mode. faces,” in Proceedings SIGCHI'99: Human Factors in An interesting observation about this Web scrolling task in Computing Systems. 1999. Pittsburgh, PA: pp. 56-63. general is that it primarily tests scrolling while searching 2. Buxton, W., “Lexical and Pragmatic Considerations of Input for information, so the scrolling must go slow enough so Structures.” Computer Graphics, 1983. 17(1) pp. 31-37. the subjects can see the content go by. This is why the 3. Buxton, W. and Myers, B. “A Study in Two-Handed Input,” in methods that provided the best control over the speed are Proceedings SIGCHI'86: Human Factors in Computing Systems. preferred. The low rating of the rate scroller on the PDA is 1986. Boston, MA: pp. 321-326. because the fastest speed was much too fast to see the text 4. Card, S.K., Moran, T.P., and Newell, A., “The Keystroke- go by, and the medium and slow speeds were rated as too Level Model for User Performance Time with Interactive Sys- slow. However, other scrolling tasks, such as those tested tems.” Communications of the ACM, 1980. 23(7) pp. 396-410. by [3], require the user to go to a known place in the July. document, and then a method that can move long distances 5. Card, S.K., Moran, T.P., and Newell, A., The Psychology of very quickly may be desirable. Human-Computer Interaction. 1983, Hillsdale, NJ: Lawrence Erlbaum Associates. CONCLUSIONS 6. Guiard, Y., “Asymmetric Divison of Labor in Human Skilled Many studies have shown the effectiveness of two-handed Bimanual Action: The Kinematic Chain as a Model.” Journal of input to computers in certain tasks. One hindrance to two- Motor Behavior, 1987. 19(4) pp. 486-517. handed applications has been that there may be only a few 7. Hinckley, K., et al., “Two-Handed Virtual Manipulation.” tasks in which using both hands is beneficial, and the ACM Transactions on Computer Human Interaction, 1998. 5(3) benefits are relatively minor. Another problem is that al- pp. 260-302. Sept. though it is interesting to study custom devices for use by 8. Jacob, R.J.K., et al., “Integrality and Separability of Input De- the non-dominant hand, in order for there to be wide-scale vices.” ACM Transactions on Computer-Human Interaction, use, it is better to provide mechanisms that users can easily 1994. 1(1) pp. 3-26. get and configure. Since increasing numbers of people 9. Kurtenbach, G., et al. “The Design of a GUI Paradigm based have PDAs that are easy to connect to PCs, it makes sense on Tablets, Two-hands, and Transparency,” in Proceedings, to see if PDAs can be used effectively in the non-dominant CHI'97: Human Factors in Computing Systems. 1997. Atlanta, hand. The research presented here shows that PDAs can be GA: ACM. pp. 35-42. used as buttons and scrollers, and that the time to home to 10. Myers, B.A., et al., “Individual Use of Hand-Held and two devices is only slightly longer than for one. Our study Desktop Computers Simultaneously,” 1999. Submitted for Publi- of one application shows that at least for the scrolling task, cation. a PDA can match or beat other 1-handed and 2-handed 11. Myers, B.A., Stiel, H., and Gargiulo, R. “Collaboration Us- techniques. Because there is no incremental cost for the ing Multiple PDAs Connected to a PC,” in Proceedings PDA since users already own it, and since the PDA is con- CSCW'98: ACM Conference on Computer-Supported Coopera- nected to the PC anyway, even small efficiencies many be tive Work. 1998. Seattle, WA: pp. 285-294. sufficient to motivate its use as a device for the non- 12. Rekimoto, J. “A Multiple Device Approach for Supporting dominant hand. Our studies and many others have empha- Whiteboard-based Interactions,” in Proceedings SIGCHI'98: sized the importance of the pragmatics and the exact Human Factors in Computing Systems. 1998. Los Angeles, CA: pp. 344-351. behavior of controls. Because the PDA can be programmed with a variety of controls with various properties, further 13. Rutledge, J. and Selker, T. “In-Keyboard Analog Pointing research is required to determine the most effective ways Device: A Case for the Pointing Stick,” in Technical Video Pro- gram of the CHI'90 Conference. 1990. SIGGRAPH Video that a PDA can be used to control the PC in both the Review, Issue 55, No. 1. dominant and non-dominant hand. 14. TestedOK Software, “Speed Typing Test 1.0. Available from ACKNOWLEDGMENTS http://hometown.aol.com/tokfiles/typetest.html. Test- For help with this paper, we would like to thank Rob Miller, [email protected],” 1999. Bernita Myers and Shumin Zhai. 15. Zhai, S., Smith, B.A., and Selker, T. “Improving Browsing Performance: A Study of Four Input Devices for Scrolling and The research reported here is supported by grants from DARPA, Microsoft, Pointing,” in Proceedings of Interact97: The Sixth IFIP Confer- IBM and 3Com. This research was performed in part in connection with Contract number DAAD17-99-C-0061 with the U.S. Army Research Labo- ence on Human-Computer Interaction. 1997. Sydney, Australia: ratory. The views and conclusions contained in this document are those of the pp. 286-292. authors and should not be interpreted as presenting the official policies or position, either expressed or implied, of the U.S. Army Research Laboratory or the U.S. Government unless so designated by other authorized documents.